Computer Networks teaches how data moves between machines; Principles of Communications explains how it moves over a physical medium: how a signal is carried on a wire, a radio wave or a fibre. It is the “physical layer” in depth. For a DevOps engineer, it explains Wi-Fi speed, bandwidth versus latency, why signal quality matters and how a link can be fast yet still lose data. This article revisits the five units of the JNTUK course.
Unit 1: Introduction and amplitude modulation
The communication system
Every system has the same blocks:
| Block | Role |
|---|---|
| Source and transducer | Information (voice, video, data) converted to an electrical signal |
| Transmitter | Prepares the signal, mainly by modulation |
| Channel | The medium: cable, fibre, free space. It adds noise and attenuation |
| Receiver | Demodulates and recovers the message |
| Destination | The user or device |
Why modulate
A voice signal (about 300 Hz to 3.4 kHz) cannot travel far as it is. Modulation shifts it onto a high-frequency carrier so that:
- antennas can be a practical size (antenna length is a fraction of the wavelength),
- many signals can share a medium on different frequencies (multiplexing),
- the signal is more resistant to noise.
Amplitude modulation (AM)
In AM, the carrier’s amplitude varies with the message: s(t) = Ac [1 + m cos(2πfmt)] cos(2πfct), where m is the modulation index. If m exceeds 1 the signal is over-modulated and distorts.
| Property | AM (full carrier) |
|---|---|
| Spectrum | Carrier plus an upper and a lower sideband |
| Bandwidth | 2 fm (twice the message bandwidth) |
| Power | Pt = Pc (1 + m²/2); only the sidebands carry information, so at m = 1 at most one third of power is useful |
| Efficiency | m² / (2 + m²) |
Variants save power and bandwidth:
| Type | What is sent | Notes |
|---|---|---|
| DSB-SC | Two sidebands, no carrier | More efficient; needs coherent detection |
| SSB | One sideband only | Half the bandwidth, less power; used in radio |
| VSB | One sideband and a trace of the other | Used for analog TV |
Transmitters use modulators (square-law, switching); receivers use an envelope detector for AM, or the superheterodyne design, which converts every station to a fixed intermediate frequency for easy filtering.
Today: AM is simple and still used in broadcast radio and aircraft communications. The idea that information rides on a carrier is the basis of everything wireless, and “modulation” in Wi-Fi and 5G is its descendant.
Unit 2: Angle modulation
In angle modulation the carrier’s amplitude stays constant while its phase or frequency carries the message.
| Type | What varies | Instantaneous frequency |
|---|---|---|
| Frequency modulation (FM) | Carrier frequency, in proportion to the message amplitude | fc + kf m(t) |
| Phase modulation (PM) | Carrier phase, in proportion to the message | Related to the derivative of m(t) |
The modulation index β = Δf / fm, where Δf is the frequency deviation. By Carson’s rule, the FM bandwidth is about 2 (Δf + fm). Narrowband FM (small β) behaves like AM in bandwidth; wideband FM (large β) uses much more bandwidth, and that extra bandwidth buys better noise immunity (the FM improvement, or “capture effect”).
| Point | AM | FM |
|---|---|---|
| Noise immunity | Poor | Good |
| Bandwidth | Narrow | Wide |
| Circuit complexity | Simple | More complex |
| Typical use | Medium-wave radio | FM radio (88 to 108 MHz), TV sound, two-way radio |
FM is generated directly (a voltage-controlled oscillator) or indirectly (Armstrong method: phase modulation followed by frequency multiplication) and detected with a frequency discriminator or a phase-locked loop (PLL). FM receivers use pre-emphasis at the transmitter and de-emphasis at the receiver to reduce high-frequency noise.
Today: FM radio, many wireless microphones and the frequency-shift keying inside simple IoT radios all use these ideas. The trade of bandwidth for noise immunity is a recurring theme that returns in Unit 5.
Unit 3: Pulse modulation and digitization
Sampling
The sampling theorem (Nyquist) says a signal limited to a maximum frequency fm can be recovered exactly if sampled at at least 2 fm samples per second. Sampling below that causes aliasing, where high frequencies appear as false low ones. Telephone speech (up to about 4 kHz) is sampled at 8 kHz; CD audio (up to 20 kHz) at 44.1 kHz.
Pulse modulation family
| Scheme | What is varied | Type |
|---|---|---|
| PAM (amplitude) | Pulse height | Analog |
| PWM / PDM (width / duration) | Pulse width | Analog |
| PPM (position) | Pulse position | Analog |
| PCM (code) | A binary number per sample | Digital |
| Delta modulation (DM) | A single bit saying “up or down” | Digital |
Pulse code modulation (PCM)
PCM is the basis of digital voice and audio. Its three steps:
- Sampling at the Nyquist rate or above.
- Quantization: round each sample to one of 2ⁿ levels, which introduces quantization noise. Each extra bit improves the signal-to-noise ratio by about 6 dB (SNR ≈ 6n + 1.8 dB).
- Encoding: write each level as an n-bit binary word.
Telephone PCM uses 8 bits at 8 kHz, so a voice channel is 64 kbit/s. Companding (µ-law in North America, A-law in India and Europe) compresses loud signals before quantizing, to give quiet sounds more precision. Delta modulation uses one bit per sample but suffers slope overload and granular noise, which adaptive delta modulation reduces. Time-division multiplexing (TDM) interleaves several PCM channels onto one link; the telephone T1/E1 hierarchy does this (E1 carries 30 voice channels at 2.048 Mbit/s).
Today: every voice call, music file and video is digitized this way. VoIP codecs (G.711 is plain 64 kbit/s PCM) and audio formats trace to PCM, and the Nyquist rule explains why 44.1 kHz was chosen for CDs and why monitoring tools can miss short spikes if they sample too slowly.
Unit 4: Digital modulation and noise
Digital modulation
To send bits over a radio or cable channel, the carrier’s properties are changed by the data.
| Scheme | What changes | Notes |
|---|---|---|
| ASK (amplitude-shift keying) | Amplitude | Simple; sensitive to noise. On-off keying is the simplest case |
| FSK (frequency-shift keying) | Frequency | More noise-robust; used in old modems |
| PSK (phase-shift keying) | Phase | Efficient. BPSK has 2 phases (1 bit per symbol), QPSK has 4 (2 bits per symbol) |
| QAM (quadrature amplitude) | Amplitude and phase together | 16-QAM carries 4 bits per symbol, 64-QAM 6, 256-QAM 8 |
Using more states per symbol (higher-order modulation) gives higher data rate in the same bandwidth but needs a cleaner signal. This is why a Wi-Fi link slows down as you move away from the router: the system falls back from 256-QAM to lower orders. A constellation diagram plots the possible symbols. OFDM splits a high-rate stream across many slow subcarriers and is used in Wi-Fi, 4G, 5G and DSL.
Noise and error performance
| Noise | Source |
|---|---|
| Thermal (Johnson) | Random motion of electrons in components; power = kTB |
| Shot | Discrete nature of current |
| Interference and crosstalk | Other signals |
The key measure is the signal-to-noise ratio (SNR), often in decibels: SNR(dB) = 10 log₁₀(S/N). The bit error rate (BER) falls steeply as the ratio of energy per bit to noise (Eb/N0) rises. A matched filter maximizes SNR at the receiver, and inter-symbol interference (ISI) is reduced with pulse shaping such as raised-cosine filters, as shown by the eye diagram. Line codes (NRZ, RZ, Manchester) decide how bits are written as voltages.
Today: the “modulation and coding scheme (MCS)” in a Wi-Fi report, a 4G/5G signal-quality reading and the “SNR margin” on a DSL or fibre line all come from this unit. Ethernet uses line codes, and a rising error counter on a switch port is often a noise or cable problem.
Unit 5: Information theory and error-control coding
Information theory
Claude Shannon’s 1948 theory measures information. The information in an event of probability p is −log₂ p bits, so rare events carry more. Entropy is the average information per symbol:
H = −Σ pᵢ log₂ pᵢ bits/symbol
It is the minimum average number of bits needed to encode the source. Source coding removes redundancy: Shannon-Fano and Huffman coding assign shorter codes to more probable symbols, and Lempel-Ziv adapts to the data (the basis of ZIP and gzip).
Channel capacity
The Shannon-Hartley theorem gives the maximum error-free rate of a noisy channel:
C = B log₂ (1 + S/N) bits per second
where B is bandwidth in Hz and S/N is the (linear) signal-to-noise ratio. For example, a 20 MHz channel with 30 dB SNR (S/N = 1000) has C ≈ 20 × 10 ≈ 200 Mbit/s. Capacity rises with bandwidth and with SNR, but only slowly with SNR; no real system can beat it.
Error-control coding
Adding redundancy lets the receiver detect or fix errors.
| Technique | What it does |
|---|---|
| Parity bit | Detects a single-bit error |
| Hamming code | Corrects single-bit errors (used in ECC memory) |
| Linear block codes | (n, k) codes with a generator matrix and syndrome decoding |
| Cyclic codes and CRC | Detect burst errors with a polynomial remainder; used in Ethernet and storage |
| Convolutional codes (Viterbi decoding) | Redundancy spread over a stream; used in older mobile and satellite links |
| Modern codes | Turbo, LDPC and polar codes, which approach Shannon’s limit; used in 4G, 5G, Wi-Fi |
Two recovery strategies: forward error correction (FEC) fixes errors at the receiver; ARQ (automatic repeat request) detects an error and asks for a resend, as TCP does.
Today: gzip, Brotli and zstd compress data using these entropy ideas. CRC32 is in Ethernet frames, ZIP files and Git, ECC memory in servers uses Hamming-style codes, and RAID parity and erasure coding in object stores such as S3 are error-correcting codes. The Shannon formula explains why Wi-Fi 6 and 5G use wider channels and more antennas rather than just more power.
One-minute recap
| Unit | Key idea | You meet it in |
|---|---|---|
| 1. AM | Message rides on a carrier; sidebands, bandwidth 2 fm | Broadcast radio, the idea of modulation |
| 2. FM and PM | Frequency or phase varies; Carson’s rule; noise immunity | FM radio, simple IoT radios |
| 3. Pulse modulation | Nyquist 2 fm; PCM = sample, quantize, encode | Voice calls, audio, VoIP codecs |
| 4. Digital modulation and noise | ASK, FSK, PSK, QAM, OFDM; SNR and BER | Wi-Fi MCS rates, 5G, DSL margin |
| 5. Information theory and coding | Entropy, C = B log₂(1 + S/N), CRC and Hamming | gzip, Ethernet CRC, ECC memory, erasure coding |
Next in Engineering Revisited: Software Engineering.
Written by Claude, Anthropic’s AI assistant, as part of Engineering Revisited, a quick refresher of the B.Tech subjects one at a time.